When customers feel the need for speed and size, Microsoft announced at Ignite a new service called Azure SQL Database Hyperscale. This service offering is now available public preview as of October 1, 2018. Think of this as cloud-based Oracle RAC for SQL Server or Amazon Aurora for SQL Server in terms of the ability to scale SQL Server up to 100 TB! Like Azure SQL Data Warehouse, Microsoft separated the Compute processing from the Storage so that you can dial up the performance needed. In this blog post, we will introduce you to the top 3 Azure SQL Database Hyperscale benefits.
1. Big size
One of the major Azure SQL Database Hyperscale benefits is that Microsoft designed it for very large databases. Microsoft tested Hyperscale at the size of 100 terabytes, and that’s where the limit comes from. Generally speaking, you can consider Azure SQL Database Hyperscale as an unlimited database. Amazon Aurora offers you just 64 terabytes of data storage, and Hyperscale goes well beyond that. Hyperscale stores the data in data pages, which are managed by the page servers.
Every page server stores up to 1 terabyte of data and acts as an independent database with its own compute power and duplicate replicas. Hyperscale adds new page servers when the available space in existing page servers is lower than a threshold. In addition, we should mention that every page server consists of 2 virtual machines for hot failover.
Despite its huge size, Azure SQL Database Hyperscale remains a regular SQL Server database in terms of the SQL code. You don’t need to update your existing queries with new syntax or consider any restrictions. You can leave your database applications as is after migrating your workloads to Azure SQL Database Hyperscale.
Managing very large databases, you typically face the following issues:
- backups significantly decrease the performance of your database applications
- restore takes quite a lot of time — not just hours, but days
- scaling of your on-premises hardware requires significant investments and fast response
Azure SQL Database Hyperscale addresses all these issues. Most importantly, backups in Hyperscale are nearly instantaneous. Likewise, Hyperscale copies the file snapshot for an instant restore. All this means that it is big not only in terms of data storage but in terms of performance too.
2. Fast performance
Faster and more reliable data throughput is another example of Azure SQL Database Hyperscale benefits which businesses can leverage. Regardless of the database size, Azure SQL Database Hyperscale operates with low latency and allows for high throughput. Microsoft promises transaction throughput at 100 MBps and constant time read scale. Another cool example from the hands-on experiments: scaling a 50 TB database from Azure SQL DB to Hyperscale takes around 2 minutes. That’s a great way to get started!
The secret of Hyperscale’s high performance lies in its new cloud-born architecture, which decouples compute, log and storage. At the same time, the processing power remains close to the data, which allows reducing network traffic and latency.
The key Hyperscale components are powered with SSDs. Also, Hyperscale stores the data in the reliable Azure storage. This brings all those cool features like 99.99% availability, disaster recovery, triple replication, and mission-critical high performance. Also, you can seamlessly tune the performance and scale.
On top of the Hypescale architecture diagram, you can see the primary read and write compute node. It is accompanied by two secondary read-only compute nodes. This solution makes scaling a very trivial process. First of all, Hyperscale scales a secondary replica and then promotes it to the primary. After that Hyperscale scales the node which previously was a primary replica.
The log service is another cool feature about the Hyperscale architecture. Hyperscale stores the log of data changes for the past 6 months, allowing for the point-in-time recovery in a couple of clicks.
3. Fair pricing
Lower costs for massive database sizes is the final example of Azure SQL Database Hyperscale benefits we’ll cover in this blog post. The Hyperscale pricing starts from $173 per month. Maintaining a live 50 TB Azure SQL Database Hyperscale will cost you a little over $300 per month. That’s definitely not an expensive offer for this kind of data storage and services that you get.
You can save an additional 40% on your license costs with Azure Hybrid Benefit program if you bring your existing SQL Server licenses.
One warning
Currently, Azure SQL Database Hyperscale runs in the public preview mode. This brand new service still lacks some of the cool features that competitors like Amazon Aurora can provide you with. Another drawback is that you can’t get back after you migrate to Hyperscale. This isn’t likely going to be a big deal once they go into general availability. However, if you find compatibility issues and need to go back to Azure SQL Database for your database – assuming it still fits – you will have to do an export and import back to Azure SQL Database.
After Microsoft solves all these issues and officially releases the Azure SQL Database Hyperscale, it will become a very strong force on the cloud computing market. Stay tuned for upcoming reviews and benchmarks of this impressive Azure cloud database service.
